Offshoring, Immigration, and the Native Wage Distribution

William W. Olney1

Department of Economics

Williams College

Abstract

This paper presents a simple model that examines the impact of offshoring and im-

migration on wages and tests these predictions using U.S. state-industry-year panel data.

According to the model, the productivity effect causes offshoring to have a more positive

impact on low-skilled wages than immigration, but this gap decreases with the workers’

skill level. The empirical results confirm both of these predictions and thus present direct

evidence of the productivity effect. Furthermore, the results provide important insight into

how specific components of offshoring and immigration affect the wages of particular types

of native workers.

Keywords: offshoring, outsourcing, immigration, productivity effect, wages

JEL Codes: F16, F22, J3

1Department of Economics, Williams College, Williamstown, MA 01267 (email:william.olney@williams.edu). I am grateful to Brian Cadena, Jennifer Hunt, Wolfgang Keller, KeithMaskus, Stephen Yeaple, and seminar participants at the 2009 Midwest International Trade Meetings,the 2009 Empirical Investigations in Trade and Geography, the 2010 Rocky Mountain Empirical TradeConference, and the University of Colorado International Trade Seminar Series for helpful comments andsuggestions.

1 Introduction

Workers in developed countries are becoming increasingly concerned about the impact of

offshoring and immigration on their domestic labor markets.2 Offshoring and immigration

are the two factors that are of most concern to American workers: 77% of Americans think

that offshoring has hurt them (13% believe it has helped) and 55% of Americans believe

immigration has hurt them (28% believe it has helped).3 While many American workers

blame their stagnant wages on the increased prevalence of offshoring and immigration, the

available evidence on the link between offshoring, immigration, and wages remains mixed.

In order to investigate the validity of these fears and clarify these relationships, this paper

presents a model that highlights the impact of offshoring and immigration on wages and

then tests these predictions using a comprehensive data set.

The offshoring of domestic jobs and the immigration of foreign workers are mechanisms

that increase the effective labor force available to domestic firms. In this respect, offshoring

and immigration will have a similar effect on wages. However, their impact on wages differs

if the benefits associated with cross country wage differences accrue to different factors of

production. Specifically, with offshoring the firm captures the rents associated with cross

country wage differences but with immigration the migrant worker captures these rents.

A simple model is constructed that clarifies these relationships between the offshoring

of low-skilled tasks, low-skilled immigration, and wages. Both offshoring and immigration

generate a labor-supply effect which depresses the wage of low-skilled workers but increases

the wage of high-skilled workers. Offshoring also generates a productivity effect which refers

to the cost savings that firms enjoy after relocating some tasks abroad. The productivity

effect increases the wage of low-skilled workers but has no direct effect on the wage of high-

skilled workers. Immigration does not generate a productivity effect since the benefits of

cross country wage differences are captured by the immigrants rather than the domestic

firms. Thus, comparing the impact of offshoring and immigration on the wages of native

workers offers a unique opportunity to test for the presence of the productivity effect.

2Offshoring refers to the relocation of domestic jobs to foreign countries.3“Public Says American Work Life is Worsening, But Most Workers Remain Satisfied with Their Jobs,”

Pew Research Center, 2006.

1

Specifically, due to the productivity effect, offshoring has a more positive impact on low-

skilled wages than immigration (Proposition 1), but this gap decreases with the workers’

skill level (Proposition 2).

The predicted impact of immigration and offshoring on the wages of different types of

native workers is then tested using a comprehensive U.S. data set that includes 48 states,

14 industries, and 7 years. This data is appealing because it exploits state and industry

variation over time and it includes service and manufacturing industries. The inclusion of

state, industry, year, state*year, and industry*year fixed effects means that factors such

as geographic differences, differences in industry productivity trends, and state macroeco-

nomic trends will be controlled for in this analysis. The endogeneity of the offshoring and

immigration decision is addressed by taking advantage of the variation in these variables

that is exogenous to local demand shocks and wages. Specifically, the offshoring instrument

is constructed using variation in offshoring that is driven by changes in foreign country

characteristics over time. The immigration instrument is constructed by taking advantage

of the fact that current immigrants often locate in areas where previous immigrants from

the same country already live (Bartel 1989). This instrumental variable estimation strategy

identifies the causal impact of offshoring and immigration on native wages.

The results confirm both propositions of the model. Offshoring of low-skilled tasks in-

creases the wages of low-skilled native workers while low-skilled immigration has a slight

negative effect on these wages. However, offshoring and immigration have a more similar

impact on the wages of high-skilled native workers. Thus, the empirical results provide

strong evidence of the productivity effect. The productivity effect is large enough to com-

pensate for the labor supply effect and causes offshoring to actually increase the wages of

low-skilled native workers. This surprising result is fully consistent with the predictions of

the model.

An additional analysis examines the impact of offshoring and immigration on native wage

deciles, rather than simply focusing on low-skilled and high-skilled wages. This provides a

more complete picture of how these forms of globalization affect the native wage distribution.

The results again confirm both predictions of the model and provide even stronger empirical

support for the productivity effect. Offshoring has a more positive impact on the wages at

2

the low end of the distribution than immigration, but this gap decreases as the wage deciles

increase.

In addition, these results provide important insight into how specific components of

offshoring and immigration affect particular types of native workers. The implications of

offshoring and immigration on wages depends on the skill level of the immigrant and on

the skill level of the offshored tasks, which is identified using the level of development of

the foreign host country. Finally, these findings are robust to using an alternate offshoring

instrument, to using a different measure of income, to changes in the sample, and to alternate

specifications.4

Some recent theoretical papers have examined the impact of offshoring and immigration

on native wages (Jones 2005, Grossman and Rossi-Hansberg 2008). These studies show

how offshoring can lead to an increase in domestic wages and discuss the similarities and

differences of immigration. This paper builds upon this literature by constructing a model

that combines immigration and offshoring into a single, unified framework. In particular,

a model is developed that incorporates immigration into a variation of the Grossman and

Rossi-Hansberg’s (2008) trade in task model. This produces specific predictions about how

offshoring and immigration affect different types of native workers. Combining offshoring

and immigration into a single framework also generates two testable predictions for the

presence of the productivity effect. This is an important contribution since it has been

diffi cult for researchers to test for the productivity effect due to the lack of adequate trade

data. The empirical results that follow support both propositions of the model and thus

provide the first direct empirical evidence of the productivity effect.

While the links between offshoring and wages (Feenstra and Hanson 1999, Slaughter

2000, Liu and Trefler 2008, Ebenstein, Harrison, McMillan, and Phillips 2009) and immi-

gration and wages (Card 1990, Borjas 2003, Card 2005, Ottaviano and Peri 2008) have been

examined extensively with results varying substantially, few studies combine offshoring and

immigration into a comprehensive empirical analysis. Not only does this provide a unique

opportunity to test for the productivity effect, it also allows for specific components of

offshoring and immigration to be compared. Conflicting results in the literature typically

4These additional results can be found in an online technical appendix.

3

arise from papers using different estimation strategies, unit of analyses, or data. However,

this paper shows that immigration and offshoring have very different impacts on native

wages depending on the skill level of immigrants and on the income level of the foreign

country. This improves our understanding of how these global forces affect the wages of

native workers and may reconcile some of the mixed results in the literature.

To the best of my knowledge, the only other papers that consider offshoring and immi-

gration in a unified framework are Ottaviano, Peri, and Wright (2010) and Barba Navaretti,

Bertola, and Sembenelli (2008).5 The first paper complements this analysis in that it fo-

cuses on the impact of offshoring and immigration on employment rather than on wages.

The second paper examines the characteristics of Italian firms that choose to offshore but

does not look at the implications of offshoring or immigration on wages. Furthermore, in

contrast to this analysis which includes a wide array of industries, both of these other papers

focus exclusively on the relatively small manufacturing sector.

Recent studies have provided highly publicized estimates of the number of U.S. jobs that

may be offshored in the coming years (Blinder 2007, Jensen and Kletzer 2005, McKinsey

Global Institute 2005). While these papers offer a rough estimate of the scope of offshoring,

they do not address the implications of offshoring for native workers. Between 22% - 29%

of all U.S. jobs are potentially offshorable (Blinder 2007), but without a clear idea of how

offshoring impacts domestic labor markets, interpreting these results is diffi cult. This paper

fills this void by identifying how different components of offshoring affect particular types of

native workers. The results that follow suggest that certain types of offshoring are beneficial

for particular types of native workers.

The remainder of the paper is organized as follows. A simple model is constructed in the

next section which highlights the impact of offshoring and immigration on wages. Section

3 describes the data used in this analysis and presents descriptive statistics while Section

4 presents the estimation strategy and describes the instrument used in the IV regressions.

The results are discussed in Section 5, and Section 6 concludes.

5 In addition, Hickman and Olney (2011) examine the impact of globalization on human capital investmentand find that community college enrollments increase in response to offshoring and immigration.

4

2 Model

Following Grossman and Rossi-Hansberg (2008), I model offshoring as trade in tasks. The

productivity effect arises in an environment in which there are heterogeneous costs of off-

shoring tasks, while the labor-supply effect arises in an environment in which there are

more factors of production than goods. Thus, in order to simply and clearly illustrate these

competing effects, the model focuses on a small economy that produces a single good using

two factors and that faces increasing costs of offshoring tasks.6 In addition, immigration,

which leads to changes in the domestic labor supply, is included in the model. While other

authors have discussed the similarities and differences of offshoring and immigration (Jones

2005, Grossman and Rossi-Hansberg 2008), this is one of the few papers that incorporates

immigration into a trade in task framework. Combining offshoring and immigration in a

unified model generates clear, testable predictions for the productivity effect.

Consider a small economy, such as a state, that takes the price and the foreign wage as

given and specializes in the production of a particular good Y. The production of good Y

requires L-workers, who are relatively less skilled, and H-workers, who are relatively more

skilled. There is a continuum of L-tasks and a continuum of H-tasks performed by each type

of worker. The tasks are defined such that each task must be performed once in order to

produce a unit of good Y. Each L-task requires aL units of domestic low-skilled labor, and

each H-task requires aH units of domestic high-skilled labor. Substitution between L-tasks

and H-tasks is possible, and thus both unit requirements are chosen by the firm in order

to minimize costs. Without loss of generality, the number of L and H tasks is normalized

to one. Therefore, aL and aH also indicate the amount of domestic L-labor and H-labor

necessary to produce a unit of good Y.

The costs associated with offshoring vary by task. This can capture a number of different

factors that can make some tasks inherently more diffi cult to offshore than others. For

6 Including a second good in the model and relaxing the small country assumption would generate, inaddition to the productivity effect and labor-supply effect, a “relative-price effect”caused by offshoring. Thiswould put downward pressure on the low-skilled wage and upward pressure on the high-skilled wage via theStolper Samuelson Theorem (Grossman and Rossi-Hansberg 2008). While this is an interesting extension,the relative price effect is not crucial for this analysis, and thus I try to keep the model as simple as possible.If anything, the relative price effect would work against the propositions of the model and the empiricalresults that follow.

5

instance, how routine or codifiable the task is, or how much face to face contact is necessary

to carry out the task, or how easy it is to electronically transmit the output can all influence

how costly it is to offshore a task (Blinder 2007). For the purposes of this analysis, we need

to simply recognize that the costs of offshoring differ by task. The L-tasks are ordered such

that the costs of offshoring are increasing and it is assumed that it is prohibitively costly to

offshore H-tasks. In addition, the immigration of L-workers to the home state is possible,

while the immigration of H-workers is negligible.7

Let w and w∗ be the wages of the L-workers in the home state and foreign country

respectively (with w > w∗). A firm can produce task j domestically at a cost of waL, or

it can produce task j abroad at a cost of w∗aLβg(j), where β is a shift parameter that

captures changes in the cost of offshoring and g(j) is a continuously differentiable function

with βg(j) ≥ 1 for all j. Firms offshore tasks in order to take advantage of lower foreign

wages but face increasing costs of offshoring, g′(j) > 0, due to the ordering of tasks. Thus,

there exists a task J such that the wage savings is exactly equal to the costs of offshoring,

or

w = βg(J)w∗. (1)

If w < βg(j)w∗, then task j is performed at home, and if w > βg(j)w∗, then task j is

performed abroad. Therefore, tasks j ∈ [0, J ], which are relatively cheap to offshore, are

carried out abroad and tasks j ∈ (J, 1], which are relatively expensive to offshore, are carried

out at home. A reduction in the cost of offshoring (dβ < 0) leads to an increase in the share

of low-skilled tasks that are offshored (dJ > 0).

If firms optimally choose aL, aH , and the tasks to offshore, then profit maximization

implies that price equals marginal cost

P = waL(.)(1− J) + w∗aL(.)

J∫0

βg(j)dj + saH(.), (2)

7While these assumptions are consistent with the findings that offshoring of high-skilled jobs and high-skilled immigration are relatively small, these restrictions will be relaxed in the empirical analysis thatfollows.

6

where s represents the high-skilled wage and aL and aH are functions of the relative average

costs of the two sets of tasks. The first term on the right-hand side represents the costs

paid to domestic low-skilled workers since (1 − J) tasks are performed at home with aL

low-skilled labor needed for each task. The second term on the right-hand side represents

the costs of hiring foreign low-skilled workers. Since the costs vary across each task, we

integrate from 0 to J . The third term is the costs of hiring native high-skilled workers.

Substituting (1) into (2) yields the following zero-profit condition:

P = Ω(J)waL(Ωw/s) + saH(Ωw/s), (3)

where

Ω(J) = 1− J +

J∫0

g(j)dj

/g(J).

Here the dependence of the factor intensities aL and aH on the relative average costs is

explicitly stated. If J = 0, then no tasks are offshored, Ω(J) = 1, and the zero-profit

condition is of the standard form. Since g′(j) > 0, by the ordering of tasks, it can be shown

that Ω(J) < 1 as long as J > 0. Therefore, the costs to the firm after offshoring some

tasks are less than if they chose to perform all L-tasks domestically. Finally, an increase

in the share of low-skilled tasks that are offshored (dJ > 0) leads to a decrease in firms’

costs (dΩ(J) < 0).8 Offshoring leads to a reduction in firms’costs through the extensive

margin because more tasks are offshored and through the intensive margin because it is

now cheaper to offshore the tasks already produced abroad.

Domestic firms reduce their costs by optimally choosing the tasks to offshore. Since

offshoring is a deliberate action on the part of the firm, offshoring features prominently

in the firms’profit maximizing decision in (3). In contrast, immigration is determined by

factors largely exogenous to the firm, such as changes in immigration policies or foreign

economic conditions. Furthermore, since domestic firms are not allowed to discriminate

8 ∂Ω∂J

=

J∫0g(j)dj

g(J)2g′(J) which is negative when J > 0. Purely for notational convenience Ω(J) will be

simplified to Ω below.

7

against immigrants by paying them lower wages, an increase in immigration does not directly

reduce firms’costs.9 Thus, immigration does not affect the profit maximizing decision facing

the firm in (3). Unlike offshoring, the benefits associated with country wage differences are

captured by the immigrants rather than the domestic firm. However, both offshoring and

immigration will have important implications for the market-clearing conditions that follow.

Each firm performs (1 − J) L-tasks at home and all H-tasks at home. Domestic firms

hire native low-skilled workers and low-skilled immigrants to perform the (1 − J) L-tasks.

Therefore, the market-clearing conditions are

(1− J)aL(Ωw/s)Y = (1 + I)N (4)

and

aH(Ωw/s)Y = H, (5)

where I ∈ [0, 1] is the ratio of immigrant low-skilled workers to native low-skilled workers

and N is the supply of native low-skilled workers. Thus, the right-hand side of (4) represents

the domestic low-skilled labor supply which consists of native and immigrant workers. H

is the supply of native skilled workers.

Using the zero profit condition and the market clearing conditions, we can examine

how an increase in offshoring or an increase in immigration affects domestic wages. Totally

differentiating equation (3), assuming that P is the numeraire, yields

θL(w + Ω) + (1− θL)s = 0, (6)

where θL is low-skilled labor’s share of total costs and hats represent a percent change.

Differentiating the ratio of (4) to (5) gives10

9As long as employers cannot fully discriminate against immigrants by paying them the prevailing wagein their source country, the cost savings under offshoring will exceed that under immigration. Furthermore,if employers can fully discriminate, then there would be no difference between offshoring and immigrationwhich would work against the empirical findings of this paper.10See Appendix for derivation.

8

σ(s− w − Ω) =dJ

(1− J)+

dI

(1 + I), (7)

where σ is the elasticity of substitution between the set of L-tasks and the set of H-tasks.

Combining (6) and (7) yields the percent change in the wage of low-skilled workers as a

function of changes in offshoring and immigration:

w = −Ω− (1− θL)

σ

dJ

(1− J)− (1− θL)

σ

dI

(1 + I). (8)

The first term on the right-hand side of (8) is the productivity effect. As the cost of

offshoring decreases (dβ < 0), more tasks are offshored (dJ > 0), and thus the cost of

performing the L-tasks declines (Ω < 0 ). Lower costs are equivalent to an increase in

productivity of low-skilled labor. Higher productivity increases the demand for low-skilled

workers and raises their wage. This is analogous to technical change that benefits low-skilled

labor. The second term on the right-hand side of (8) is the labor-supply effect of offshoring.

As the cost of offshoring decreases (dβ < 0), more L-tasks are offshored (dJ > 0), and

thus some low-skilled workers become unemployed. In order to absorb this excess supply

of workers back into the labor market, the low-skilled wage declines. Together the first and

second terms of equation (8) represent the impact of offshoring on the wage of low-skilled

workers in this model. The third term on the right-hand side of (8) is the labor-supply

effect caused by immigration. The excess supply of low-skilled workers due to immigration

reduces the low-skilled wage. From equation (8), the following proposition is immediate:

Proposition 1 Due to the productivity effect, offshoring has a more positive impact on the

wage of low-skilled workers than immigration.

While both offshoring and immigration generate a labor-supply effect, offshoring also

generates a productivity effect that increases the wages of low-skilled workers. The pro-

ductivity effect is larger when more tasks are already being done abroad (J > 0) because

the cost savings at the intensive margin will be larger. The cost savings at the extensive

margin is negligible due to the envelope theorem. The labor-supply effect is small when the

share of low-skilled labor in total costs (θL) is large or when H-tasks and L-tasks are good

9

substitutes (σ is large). Thus, under these conditions, the productivity effect will exceed

the labor-supply effect and offshoring will increase the wages of low-skilled workers. This

generates the seemingly counterintuitive result that offshoring can benefit the factor whose

tasks are being sent abroad.

Immigration, on the other hand, unambiguously decreases the wage of low-skilled labor

in this model. Immigration does not generate a productivity effect because the benefits of

country wage differences are captured by the immigrants rather than the domestic firm.

Unlike offshoring, immigration does not generate any direct cost savings for domestic firms

since they pay immigrants and native workers the same market wage.

Using (6) and (7), it is also possible to derive the percent change in the wage of high-

skilled workers as a function of changes in offshoring and immigration:

s =θLσ

dJ

(1− J)+θLσ

dI

(1 + I). (9)

Here the labor-supply effect of offshoring and immigration increases the wage of high-skilled

workers. As is common in a two factor model, an increase in the effective supply of low-

skilled labor increases the marginal product and wage of high-skilled workers. Offshoring

does not generate a productivity effect for high-skilled workers because a decrease in the

costs of offshoring (dβ < 0) reduces the firms’costs of performing L-tasks with no effect on

the costs of performing H-tasks. In other words, there is no direct impact of offshoring on

the productivity of skilled workers although offshoring can indirectly affect the skilled wage

through the labor-supply effect. Comparing equations (8) and (9) establishes the following

proposition:

Proposition 2 Due to the productivity effect, the impact of offshoring and immigration on

wages becomes more similar as the workers’skill level increases.

The labor-supply effects generated by offshoring and immigration have a negative impact

on the low-skilled wage and a positive impact on high-skilled wage. However, the productiv-

ity effect generated by offshoring only impacts the low-skilled wage since offshoring affects

the costs of performing L-tasks but not H-tasks. Thus, offshoring and immigration differ in

10

their impact on the low-skilled wage but have a similar impact on high-skilled wages.

The finding that the productivity effect only impacts the low-skilled wage is an inter-

esting and important result. The reduction in costs associated with offshoring L-tasks is

analogous to an increase in the productivity of low-skilled workers which, in a general equi-

librium context, drives up their wage. The costs associated with performing the H-tasks

is unchanged and thus offshoring does not generate a productivity effect for the skilled

wage.11

The model focuses on the offshoring of low-skilled tasks and the immigration of low-

skilled workers in part because recent experience has indicated that these are particularly

important components of globalization, especially in the U.S. However, the model can

easily be extended to include the offshoring of high-skilled tasks and the immigration of

high-skilled workers. This would not change the existing results of the model but would

rather add analogous productivity and labor-supply effects generated by skilled offshoring

and immigration that work in the opposite direction. Specifically, the offshoring of high-

skilled tasks would generate its own productivity effect that would increase the wage of

high-skilled workers but have no impact on the wage of low-skilled workers. In addition,

both skilled offshoring and skilled immigration would generate a labor-supply effect that

would depress the high-skilled wage and increase the low-skilled wage. While not the focal

point of this model, these other types of offshoring and immigration will be controlled for

in the empirical analysis.

The productivity and labor-supply effects generated by offshoring are not specific to the

assumptions of the model presented in this paper. They are quite general findings that arise

in a wide variety of different theoretical frameworks.12 However, the model outlined in this

paper is appealing because it incorporates immigration into this trade in task framework and

yet simply and clearly compares the competing productivity and labor-supply effects. Fur-

11Abstracting from the model, there are certainly other plausible explanations for who captures the savingsassociated with offshoring. For instance, capital owners, entrepreneurs, consumers, or even skilled workersmay also benefit from the cost savings generated by offshoring. However, the empirical results that followare consistent with the predictions of the model and indicate that the productivity effect increases the wagesof low-skilled workers with little impact on high-skilled wages. Additional results in the appendix suggestthat these alternate hypotheses are relatively less important.12See Grossman and Rossi-Hansberg (2008) for a discussion of the productivity effect and the labor-supply

effect in many of these different environments.

11

thermore, an especially nice aspect of this model is that it highlights a unique opportunity

to test for the productivity effect by comparing the impact of offshoring and immigration

on wages. Specifically, due to the productivity effect offshoring and immigration should

have a different impact on the wage distributions within an industry. The remainder of the

paper examines whether there is empirical evidence that wages respond to offshoring and

immigration in the manner predicted by Propositions 1 and 2.

3 Data

Using state, industry, and year variation, the empirical analysis that follows will test these

predictions by estimating the impact of offshoring and immigration on the wages of U.S.

native workers. Specifically, each state-industry labor market is characterized by the econ-

omy discussed in the model and differences across states, industries, and years are used to

identify the impact of offshoring and immigration on the native wage distribution. Thus, the

data set utilized in this analysis spans the 48 contiguous U.S. states, 14 NAICS industries,

and 7 years (2000-2006).

3.1 Immigration and Offshoring Variables

Census and American Community Survey data on employed individuals who earn a positive

wage, are not in school, and are between the ages of 18 and 65 is obtained from the Integrated

Public Use Microdata Series (IPUMS). From these individual census observations, native

wage percentiles are constructed for each state-industry-year observation.

Low-skilled immigration is calculated as the share of employed individuals who are

foreign born and have a high school degree or less. This is consistent with I from the

model. In addition, the following control variables are calculated for each observation: the

share of high-skilled immigrants, the share of native employees that are male, the share of

native employees that are of a particular race and marital status, and the average age and

average educational attainment of native workers.

Data on offshoring, defined as the number of employees at majority-owned foreign affi l-

12

iates of U.S. firms, is obtained from the U.S. Bureau of Economic Analysis (BEA).13 Given

the trade in task model, focusing on foreign affi liate employment is preferable to other mea-

sures of foreign direct investment such as affi liate sales. The BEA provides foreign affi liate

employment data by year, industry, and country of the foreign affi liate.

The empirical analysis focuses on the offshoring of low-skilled tasks to less-developed

countries.14 This is consistent with the model in two important ways. First, offshoring to

less-developed countries typically entails relocating particular production tasks abroad in

order to take advantage of low foreign wages. This vertical offshoring is exactly the type

of offshoring modeled in the trade in task framework. In contrast, offshoring to developed

countries typically entails relocating the entire production process abroad in order to avoid

transport costs. Second, offshoring to less-developed countries typically, although not al-

ways, entails the offshoring of low-skilled tasks. This is consistent with the assumption of

the model that the offshoring of L-tasks is possible while the offshoring of H-tasks is pro-

hibitively expensive. Thus, offshoring to less-developed countries is most consistent with

the type of offshoring envisioned in the model.

Since offshoring data is not available by state, foreign affi liate employment (FAE) is

distributed across states based on the share of state GDP to national GDP in that industry.

Given the potential correlation between GDP and wages, the pre-sample 1999 GDP shares

are used to construct the offshoring measure. Finally, the share of foreign affi liate employ-

ment to total employment, including both domestic and foreign employment, is calculated

by state, industry, and year. Thus, offshoring is defined as the following share

13While the model does not draw a distinction between offshoring tasks to foreign affi liates or foreign arms-length suppliers, the empirical section of this paper will focus on the offshoring of jobs to foreign affi liatesdue to data constraints. Constructing a proxy of offshoring that includes arms-length suppliers (Feenstraand Hanson 1999) is only possible in the manufacturing sector which would severly limit the sample. Sinceoffshoring to arms-length suppliers is diffi cult to measure, and given that offshoring to foreign affi liates isrelatively less labor intensive (Antras 2003), this definition represents a lower bound on the total amount ofoffshoring.14The BEA does not provide information on the skill level of foreign affi liate employees. Thus, low-skilled

offshoring is measured using foreign affi liate employment in less-developed countries. This is calculated asthe difference between total foreign affi liate employment and foreign affi liate employment in Europe, Canada,Australia, and Japan.

13

L_offsit =

[GDPsi1999∑sGDPsi1999

∗ FAE_lessdevit]

Domestic_Emplsit +

[GDPsi1999∑sGDPsi1999

∗ FAEit] ∗ 100,

where s indexes states, i indicates industries, and t references years. This measure of off-

shoring is consistent with J from the model which captures the share of L-tasks that are

offshored. Offshoring to developed countries and inshoring, defined as the number of em-

ployees at majority-owned U.S. affi liates of foreign firms, were constructed in an analogous

manner and will be important control variables. Comparing total offshoring to data from

the Trade Adjustment Assistance (TAA) program indicates that this method of distributing

foreign affi liate employment across states is accurate.15

3.2 Appealing Aspects of the Data

Despite the limitations of the BEA data, this data set has a number of appealing features.

First, using U.S. state level data is preferable to a cross country analysis where it is diffi cult

to control for unobserved factors. Since U.S. states share similar laws, institutions, and

cultural characteristics, using states as the unit of analysis limits these confounding factors.

Together with the variation in offshoring and immigration across states, this means that the

link between these forms of globalization and wages is more easily identified. In addition,

state level data mitigates many of the mobility concerns associated with a city or county

level study. Thus, states more closely resemble a closed labor market while still offering a

substantial amount of variation.

Second, this analysis incorporates 14 2-digit NAICS industries which range from man-

ufacturing to professional services to finance.16 Due to data constraints, many previous

studies focus just on manufacturing industries (Feenstra and Hanson 1999, Harrison and

McMillan 2006, Amiti and Wei 2009, Ottaviano et al. 2010). However, manufacturing rep-

15The TAA program has data on the number of domestic workers who are displaced due to importcompetition. While these variables measure slightly different things, the correlation coeffi cient betweenthese two variables is 0.8.16Available BEA data on foreign affi liate employment restricts the analysis to these 14 industries. More

disaggregated foreign affi liate employment data by industry-country-year has many more missing valuesdue to confidentiality concerns. "Agriculture" and "mining" were combined and "professional services" and"management" were combined due to a lack of census observations by state in these industries.

14

resents only 13% of total U.S. GDP in 2008.17 Unlike these previous studies which study

a small component of the U.S. economy, this analysis examines how offshoring and immi-

gration affect wages in a wide variety of industries. Furthermore, by focusing on highly

aggregated NAICS industries and a relatively short time period, mobility across industries

is less problematic. In addition, any potential bias associated with mobility across industries

or states would attenuate the results.

Finally, the years included in this analysis span exogenous shocks to both offshoring and

immigration caused by China joining the World Trade Organization in 2001 and changes

to immigration policy following 9/11.

3.3 Descriptive Statistics

Figure 1 plots average immigration and offshoring by state. Not surprisingly, the urban

coastal states of California, New York, and New Jersey have high shares of both, Florida

and Nevada have high shares of immigration, midwestern rust-belt states such as Michigan

and Indiana have a high shares of offshoring, and rural isolated states such as Montana

and North Dakota have low shares of both. This variation in offshoring and immigration is

consistent with anecdotal evidence and provides additional confirmation that the allocation

of foreign affi liate employment across states is accurate.

Figure 2 plots average immigration and offshoring by industry. Not surprisingly, there

is a lot of offshoring in manufacturing but relatively little in construction, real estate,

and health care. There are high immigrant shares in accommodations, administration, and

agriculture while utilities, finance, and information have relatively little immigration. The

substantial variation in wages, immigration, and offshoring across states and industries

supports the assertion that a state-industry labor market is reasonably closed. Although

the fixed effects will capture much of the variation in Figures 1 and 2, these figures provide

insight into the dimensions and nature of the data set used in this analysis.

To gain a sense of the variation exploited in this analysis, I need to eliminate the variation

that will be captured by the various fixed effects. This is done by first regressing the median

native wage, offshoring, and immigration variables on year, state, industry, state*year, and

17Gross Domestic Product by Industry Accounts (BEA).

15

industry*year fixed effects. The residuals from these regressions will be the variation left

after accounting for the fixed effects which is the focus of this analysis. The median wage

residuals and offshoring residuals are plotted in Figure 3 while the median wage residuals

and immigration residuals are plotted in Figure 4. It is evident in Figure 3 that offshoring is

associated with a higher median native wage. However, there is little relationship between

immigration and the median native wage in Figure 4. These basic scatter plots suggest

that there is an important difference between the impact of offshoring and immigration on

native wages. However, to more accurately test the propositions of the model, it is necessary

to examine the causal impact of low-skilled offshoring and low-skilled immigration on the

wages of different types of native workers.

4 Estimation Strategy

4.1 Specification

To test the propositions of the model, the empirical analysis examines the key relationships

of the model specified in equations (8) and (9). Specifically, the impact of offshoring and

immigration on low-skilled wages is estimated by adding an error term, εsit, to (8). The error

term captures unobserved factors not explicitly included in the model that may influence

wages. This generates the following estimation equation

L_wagesit = α0 + α1L_Offsit + α2L_Imgsit + εsit,

where s indexes states, i indexes industries, t indexes years, and L_wage is the ln of

the 25th percentile native wage. L_Off is offshoring of low-skilled tasks and will capture

the productivity and labor-supply effects generated by offshoring (the first two terms on the

right hand side of (8)). L_img is low-skilled immigration and will capture the labor-supply

effect generated by immigration (the third term on the right hand side of (8)). It is assumed

that εsit takes the following form

εsit = α′3Xsit + δs + ηi + γt + ψst + νit + εsit,

16

where X is a set of control variables, δs are state fixed effects, ηi are industry fixed

effects, γt are year fixed effects, ψst are state*year fixed effects, νit are industry*year fixed

effects, and εsit is measurement error. The X matrix includes other globalization variables,

such as offshoring to developed countries, high-skilled immigration, and inshoring, as well

as demographic controls of the native population. The inclusion of state, industry, and

year fixed effects means that factors such as geographic differences, differences in indus-

try productivities, and macroeconomic factors will be controlled for in this analysis. The

state*year and industry*year fixed effects will capture trends such as growth in states over

time and changes in industry productivity over time which may be correlated with wages,

immigration, and offshoring.18

Applying a similar transformation to (9) generates the two key estimation equations:

L_wagesit = α0 +α1L_Offsit+α2L_Imgsit+α′3Xsit+δs+ηi+γt+ψst+νit+εsit, (10)

H_wagesit = ρ0 +ρ1L_Offsit+ρ2L_Imgsit+ρ′3Xsit+ δs+ηi+γt+ψst+νit+ εsit, (11)

where H_wage is the ln of the 75th percentile native wage. Proposition 1 of the model

predicts that α1 > α2 while Proposition 2 of the model predicts that ρ1 = ρ2.

While consistent with the model, estimating the impact of offshoring and immigration

on the 25th and 75th wage percentiles may miss important implications for other parts of

the native wage distribution. To gain greater insight into the impact on the entire wage

distribution, the effect of offshoring and immigration on different wage deciles of native

workers is examined. Furthermore, this is especially useful in testing Proposition 2 of the

model. Thus, the following equation will be estimated:

18Unfortunately, there is not enough annual variation to include state*industry, state*year, and indus-try*year fixed effects.

17

wagesitd = α0 +α1L_Offsit +α2L_Imgsit +α′3Xsit + δs + ηi + γt +ψst + νit + εsitd. (12)

where d indexes native wage deciles. Separate regressions are run for each native wage decile.

The model predicts that due to the productivity effect α1 > α2 for low wage deciles but as

the native wage deciles increase the productivity effect diminishes and thus the difference

between α1 and α2 decreases.19

4.2 Instruments

The inclusion of a wide range of control variables and numerous fixed effects limits the

endogeneity problem. However, a remaining potential issue is that an industry-state spe-

cific labor demand shock may be correlated with wages, immigration, and offshoring. For

instance, an increase in the productivity of low-skilled workers could increase the demand

for these workers and drive up their wage, which may encourage immigration and provide

an incentive to offshore. To address these endogeneity concerns, this paper constructs in-

struments for the offshoring and immigration variables. Essentially these instruments use

the variation in offshoring and immigration that is driven by factors in foreign countries

which is exogenous to local demand shocks and wages.

The offshoring instrument is constructed by first regressing the foreign affi liate employ-

ment data on industry-year and country-year fixed effects.20 The country-year coeffi cients

are then used to construct the instrument while the industry-year coeffi cients are discarded.

The concern is that the industry-year variation may be driven in part by local demand shocks

that could be correlated with wages. In contrast, the country-year variation is driven by

changes in foreign country characteristics over time. Percent changes in the country-year

coeffi cients are then multiplied by the pre-sample 1999 level of offshoring across industries

in each country. This generates a predicted measure of U.S. offshoring that is driven by

19Quantile regressions are not used because the goal of this analysis is to examine the impact of offshoringand immigration on the overall wage distribution not on the wage distribution conditional on the controlvariables (essentially the distribution of the error term). Furthermore, the different levels of aggregationbetween wages, which would need to be at the individual level, and the globalization variables, which are atthe state-industry level, make quantile regressions impossible.20Ottaviano, Peri, and Wright (2010) use a similar method to calculate their "imputed offshoring" measure.

18

changes in the foreign country and that is independent of domestic factors. Finally, the

predicted measure of foreign affi liate employment is aggregated and attributed to states

based on the methodology discussed in section 3.1.

The immigration instrument takes advantage of the fact that immigrants often settle

in areas where previous immigrants from the same country already live (Bartel 1989).21

Specifically, the instrument is constructed by assigning actual immigrants in the current year

to state-industries where immigrants from the same region of the world were located in the

initial year.22 The sum of immigrants across the regions is then divided by total employment

to get the predicted immigrant share for each skill group. This instrument captures variation

in immigration that is due to factors in the foreign country that influence the total number of

immigrants and which will affect states and industries differently depending on the initial

distribution of immigrants. Thus, the variation in immigration that is driven by labor

demand factors and that may be endogenous to wages is eliminated.

To be valid, these offshoring and immigration instruments need to be correlated with

the potentially endogenous variables and uncorrelated with the error term, ε, in the sec-

ond stage. While the former condition is easily tested using the first stage F-stat of the

instruments, the latter is not since the over-identification test is not possible given that the

number of instruments equals the number of endogenous variables. However, it is likely

that the exclusion restriction is satisfied since variation in offshoring and immigration that

is driven by changes in foreign country characteristics is unlikely to be correlated with un-

observed factors at the industry-state level that affect wages. This is even more plausible

given the inclusion of the state, industry, year, state*year, and industry*year fixed effects.

21This instrument is similar to the one used by Card (2001), Cortes (2008), and Peri and Sparber (2009).Ottaviano, Peri, and Wright (2010) construct a similar "imputed immigration" measure that uses differencesin the presence of immigrant groups across industries.22The eight regions of the world are U.S. Areas, Canada, Central America, South America, Europe, Asia,

Africa, and Oceania.

19

5 Results

5.1 Low-Skilled and High-Skilled Wages

The OLS results of estimating (10) and (11) are reported in the first two columns of Table

1. All regressions are weighted by the sample size, include the full set of fixed effects, and

have robust standard errors in brackets. The results are interesting and provide preliminary

support for the predictions of the model. Low-skilled offshoring has a more positive impact

on the low-skilled native wage than on the high-skilled native wage. Specifically, a one

percentage point increase in the share of offshoring increases the low-skilled wage 4.4%

and increases the high-skilled wage by 1.0%. Low-skilled immigration has a slight negative

impact on both the low-skilled and high-skilled wage.

The coeffi cients on these two key variables of interest are consistent with the predictions

of the model. Offshoring has a more positive impact on low-skilled wages than immigration

which supports Proposition 1 of the model. However, offshoring and immigration have a

relatively similar impact on high-skilled wages which supports Proposition 2 of the model.

Furthermore, the magnitude and sign of the coeffi cients indicate that the productivity is

important but that the labor-supply effect is small.

Columns three and four in Table 1 report the IV results from estimating (10) and

(11). The IV results indicate that a one percentage point increase in offshoring increases

the low-skilled wage 4.8% while a one percentage point increase in immigration decreases

the low-skilled wage 0.2%. A one percentage point increase in offshoring increases the

high-skilled wage 1.6% while a one percentage point increase in immigration decreases the

high-skilled wage 0.1%. Again, offshoring has a more positive impact on low-skilled wages

than immigration but this gap decreases with the workers’ skill level. The productivity

effect is empirically important at the low end of the wage distribution but as it dissipates

the impact of offshoring and immigration on wages becomes more similar.

While not the focal point of this analysis, the coeffi cients on the other globalization mea-

sures are interesting. Offshoring to developed countries is measured by the H_Offshoring

variable and typically entails replicating the production process abroad in order to access

foreign markets and save on transport costs. Since foreign workers are substituting for

20

domestic workers it is not surprising that H_Offshoring depresses native wages in the

IV regressions in Table 1. We also see that high-skilled immigration has a slight negative

impact on the wages of high-skilled native workers. Not surprisingly inshoring increases

native wages.

Comparing the OLS and IV results in Table 1 indicates that the OLS coeffi cients on

L_Offshoring are biased down while the coeffi cients on L_Immigration,H_Offshoring,

and H_Immigration are biased up. This suggest that a positive local demand or produc-

tivity shock, which increases wages, decreases offshoring to less developed countries but

increases offshoring to developed countries. The positive bias in the immigration coeffi -

cients indicates that immigrants, particularly relatively mobile high-skilled immigrants, are

attracted to high wage states and industries. While there is some evidence of endogeneity

bias in the OLS regressions, overall the main findings in both the OLS and IV results are

consistent with the propositions of the model.

Table 2 reports the first stage IV results. Each of the four instruments have a strong

impact on the globalization variable it was designed to predict. All of these coeffi cients

are positive and significant at the one percent level. The first stage F-stat on the excluded

instruments is above 63 in all regressions.

5.2 Wage Deciles

To gain greater insight into how offshoring and immigration affect the native wage distribu-

tion, equation (12) is estimated. Specifically, separate regressions are run using each native

wage decile as the dependent variable. The OLS results are reported in Table 3.23

Again the results support both propositions of the model. L_Offshoring has a more

positive impact on low-skilled native wages than L_Immigration but this difference de-

creases as the wage deciles increase. Specifically, the second proposition of the model, that

the impact of offshoring and immigration on wages becomes more similar as the workers

skill level increases, is verified at multiple points in the native wage distribution. The more

23Unfortunately, the Census replaces wage values above $200,000 with the state average of these wagevalues regardless of industry. While it is important to include these ‘top coded’observations in order tomaintain an accurate wage distribution, regressions using the 90th wage decile as the independent variableare biased and are therefore not reported.

21

comprehensive wage decile analysis provides even more compelling evidence in support of

the propositions of the model.

The results in Table 3 also indicate that the components of offshoring and immigration

have very different impacts on native wages. L_Offshoring increases the wages of many

native workers, particularly at the low end of the distribution, while H_Offshoring de-

creases the wages of many native workers. However, L_Immigration decreases the wages

of native workers while H_Immigration increases the wages of native workers. These con-

trasting results highlight the importance of controlling for the immigrant skill level and skill

level of offshored tasks using the level of development of the foreign country.

In addition to differences between the independent variables, there are also important

differences in how offshoring and immigration impact various types of native workers. An

appealing aspect of using native wage deciles is the ability to examine how offshoring and

immigration affect wage inequality. The results in Table 3 indicate that L_Offshoring

decreases wage inequality since the wages at the low end of the distribution increase by

relatively more than the wages at the high end. However, H_Offshoring increases wage

inequality. In contrast, immigration does not have a significant effect on wage inequality.

Table 3 indicates that both types of immigration have a relatively constant effect on the

wages of different types of native workers.

Table 4 reports the IV wage decile results. The IV analysis is appealing because it

identifies a causal impact of offshoring and immigration on native wages by exploiting

the variation in offshoring and immigration that is exogenous to local demand and wage

conditions. The coeffi cients imply that, for instance, a one percentage point increase in low-

skilled offshoring increases the median wage 3.5%, while a one percentage point increase in

low-skilled immigration decreases the median wage 0.2%. Overall, the results in Table 4

provide even stronger support for Proposition 1 and Proposition 2 of the model. Offshoring

generates a productivity effect that more than compensates for the labor supply effect at the

low end of the wage distribution. However, as the wage deciles increase, the productivity

effect diminishes, and thus the impact that offshoring and immigration have on native wages

converges.

The L_Offshoring and L_Immigration coeffi cients and their 95% confidence intervals

22

from Table 4 are plotted in Figure 5. The figure shows that offshoring has a more positive

impact on the wages of low-skilled native workers than immigration (consistent with Propo-

sition 1 of the model), but this gap decreases as the wage deciles increase (consistent with

Proposition 2 of the model). According to the model, low-skilled immigration generates

only a labor-supply effect that depresses the wages of low-skilled workers and increases the

wages of high-skilled workers. The L_Immigration coeffi cients suggest that the labor-

supply effect is relatively small. In contrast, the productivity effect, which is represented

by the vertical distance between the two lines in Figures 5, is relatively large particularly

at the low end of the distribution where the model predicts it should be strongest.

One potential concern is that the offshoring of low-skilled tasks and low-skilled immi-

gration may simply displace the least skilled, lowest wage decile native workers. As these

low-skilled native workers become unemployed, each wage decile would then capture a

slightly more educated, higher paid native worker. However, the inclusion of the average

education attainment of the native population variable, which proved to be an important

control, will account for these types of compositional shifts in employment. Furthermore,

if this compositional shift in employment was driving these results, one would observe off-

shoring and immigration leading to an initial increase in all the native wage deciles.24 The

fact that neither the L_Offshoring nor the L_Immigration coeffi cients exhibit these

patterns indicates that there is little empirical support for this hypothesis.

In contrast, the model predicts that offshoring and immigration will displace low-skilled

native workers and depress their wages via the labor-supply effect. Thus, the model is fully

consistent with this displacement hypothesis but the implications for native wages are quite

different. The fact that the empirical results provide support for this labor-supply effect

and not for the composition shift in employment story further strengthens the arguments

presented in this paper.

24Given the exponential distribution of wages, it is likely that the higher wage deciles would increase bymore than the lower wage deciles.

23

6 Conclusion

Workers have become increasingly concerned about the impact that offshoring and immigra-

tion have on wages. Despite extensive research, which generally focuses on one or the other

of these phenomena, the available evidence on the link between offshoring, immigration,

and wages remains mixed. This paper presents a simple model that identifies the ways in

which offshoring and immigration can affect wages. Both offshoring and immigration gener-

ate a labor-supply effect, while offshoring also generates a productivity effect that benefits

low-skilled native workers. Thus, comparing the impact of offshoring and immigration on

native wages offers a unique opportunity to test for the productivity effect.

The empirical results confirm that offshoring and immigration have different impacts on

native wages and highlight the importance of the productivity effect. Consistent with the

propositions of the model, offshoring has a more positive impact on low-skilled native wages

than immigration, but this difference decreases as the wage deciles increase. These results

provide direct empirical evidence that offshoring generates a productivity effect that benefits

the factor whose jobs are sent abroad. More generally the empirical results presented in this

paper improve our understanding of how offshoring and immigration affect native wages.

The findings move us past simply thinking about whether offshoring and immigration are

good or bad for the domestic economy, and instead identifies how specific components of

offshoring and immigration affect particular types of native workers.

Overall, this paper shows that the impact of these important types of globalization

on wages is not as bad as many American workers fear. Specifically, the offshoring of

low-skilled tasks to less-developed countries complements domestic workers, generates a

productivity enhancing effect, and increases low-skilled native wages. However, there is

evidence that certain components of offshoring and immigration can depress the wages of

specific types of native workers. Policy makers need to recognize these differences and take a

more nuanced approach to offshoring and immigration. Obviously the impact of offshoring

and immigration on other dimensions of the home and foreign economies are important and

warrant further research.

24

References

Amiti, Mary and Shang-Jin Wei (2009) “Service Offshoring and Productivity: Evidence

from the US,”World Economy, 32(2): 203-20

Antràs, Pol (2003) “Firms, Contracts, and Trade Structure,”Quarterly Journal of Eco-

nomics, 118(4): 1375-418

Barba Navaretti, Giorgio, Giuseppe Bertola, and Alessandro Sembenelli (2008) "Offshoring

and Immigrant Employment: Firm-level Theory and Evidence," Discussion Paper No.

6743, Center for Economic Policy Research

Bartel, Ann P. (1989) “Where Do the New U.S. Immigrants Live?”Journal of Labor Eco-

nomics, 7(4): 371-91

Blinder, Alan S. (2007) “How Many U.S. Jobs Might Be Offshorable?”Working Paper No.

142, Center for Economic Policy Studies

Borjas, George J. (1995) “Assimilation and Changes in Cohort Quality Revisited: What

Happened to Immigrant Earnings in the 1980s?” Journal of Labor Economics, 13(2):

201-45

Borjas, George J. (2003) “The Labor Demand Curve is Downward Sloping: Reexamining

the Impacts of Immigration on the Labor Market,” Quarterly Journal of Economics,

118(4): 1335-74

Card, David (1990) “The Impact of the Mariel Boatlift on the Miami Labor Market,”

Industrial and Labor Relations Review, 43(2): 245-57

Card, David (2001) “Immigrant Inflows, Native Outflows, and the Local Market Impacts

of Higher Immigration,”Journal of Labor Economics, 19(1): 22-64

Card, David (2005) “Is the New Immigration Really so Bad?” The Economic Journal,

115(507): F300-23

Cortes, Patricia (2008) "The Effect of Low-Skilled Immigration on U.S. Prices: Evidence

from CPI Data," Journal of Political Economy, 116(3): 381-422

25

Ebenstein, Avraham, Ann Harrison, Margaret McMillan, and Shannon Phillips (2009) "Es-

timating the Impact of Trade and Offshoring on American Workers Using the Current

Population Surveys," Working Paper No. 15107, National Bureau of Economic Research

Feenstra, Robert C. and Gordon H. Hanson (1999) “The Impact of Outsourcing and High-

Technology Capital on Wages: Estimates for the United States, 1979-1990,”The Quar-

terly Journal of Economics, 114(3): 907-40

Feenstra, Robert C. (2007) “Globalization and Its Impact on Labor,”Working Paper No.

44, Vienna Institute for International Economic Studies

Grossman, Gene M. and Esteban Rossi-Hansberg (2008) “Trading Tasks: A Simple Theory

of Offshoring,”American Economic Review, 98(5): 1978-97

Hanson, Gordon H., Kenneth F. Sheve, Matthew J. Slaughter, and Antonio Spilimbergo

(2002) “Immigration and the U.S. Economy: Labor-Market Impacts, Illegal Entry, and

Policy Choices.”In Immigration Policy and the Welfare System, ed. Tito Boeri, Gordon

Hanson, and Barry McCormick, 169-279 Oxford University Press

Harrison, Ann E. and Margaret S. McMillan (2008) “Offshoring Jobs? Multinationals

and U.S. Manufacturing Employment,”Working Paper, National Bureau of Economic

Research

Hickman, Daniel C. and William W. Olney (2011) "Globalization and Investment in Human

Capital," Industrial and Labor Relations Review, 64(4): 652-670

Jensen, J. Bradford and Lori G. Kletzer (2005) “Tradable Services: Understanding the

Scope and Impact of Service Offshoring,”In Brookings Trade Forum: 2005, ed. Susan M.

Collins and Lael Brainard

Jones, Ronald W. (2005) “Immigration vs. Outsourcing: Effects on Labor Markets,”Inter-

national Review of Economics & Finance, 14(2): 105-14

Liu, Runjuan and Daniel Trefler (2008) "Much Ado About Nothing: American Jobs and

the Rise of Service Outsourcing to China and India," Working Paper No. 14061, National

Bureau of Economic Research

26

McKinsey Global Institute (2005) The Emerging Global Labor Market.

Ottaviano, Gianmarco I.P. and Giovanni Peri (2008) "Immigration and National Wages:

Clarifying the Theory and Empirics," Working Paper No. 14188, National Bureau of

Economic Research

Ottaviano, Gianmarco I.P., Giovanni Peri, and Greg C. Wright (2010) "Immigration, Off-

shoring, and American Jobs," Working Paper No. 16439, National Bureau of Economic

Research

Peri, Giovanni and Chad Sparber (2009) "Task Specialization, Immigration, and Wage,."

American Economic Journal: Applied Economics, 1(3): 135-169.

Ruggles, Steven, J. Trent Alexander, Katie Genadek, Ronald Goeken, Matthew B.

Schroeder, and Matthew Sobek. Integrated Public Use Microdata Series: Version 5.0

[Machine-readable database]. Minneapolis: University of Minnesota, 2010

Slaughter, Matthew J. (2000) “Production Transfer within Multinational Enterprises and

American Wages,”Journal of International Economics, 50(2): 449-72

27

FIGURE 1Immigration and Offshoring by State

State average of the share of employees that are foreign born and theshare of employees that work abroad weighted by the sample size.

AL

AZ

AR

CA

CO

CT

DE

FL

GA

ID

IL

INIAKS

KYLAME

MD

MA

MIMN

MSMO

MT

NE

NV

NH

NJ

NM

NY

NC

NDOH

OK

OR

PA

RI

SC

SDTN

TX

UT

VT

VAWA

WVWIWY

010

2030

40Im

mig

ratio

n

2 4 6 8 10Offshoring

FIGURE 2Immigration and Offshoring by Industry

Industry average of the share of employees that are foreign born andthe share of employees that work abroad weighted by the sample size.

Agriculture

Utilities

ConstructionManufacturing

Wholesale Trade

Retail TradeTransportation

InformationFinance

Real EstateProfessional Services

Administration and Waste Services

Health Care

Accomodations and Food Services

510

1520

25Im

mig

ratio

n

0 5 10 15 20Offshoring

28

The residuals from regressing the ln native median wage on state,industry, and year fixed effects are plotted against the residuals fromregressing offshoring on state, industry, and year fixed effects.

FIGURE 3Median Wage and Offshoring

­1­.5

0.5

1M

edia

n W

age

Res

idua

ls

­10 ­5 0 5 10 15 20Offshoring Residuals

The residuals from regressing the ln median native wage on state,industry, and year fixed effects are plotted against the residuals fromregressing immigration on state, industry, and year fixed effects.

FIGURE 4Median Wage and Immigration

­1­.5

0.5

1M

edia

n W

age

Res

idua

ls

­20 ­10 0 10 20 30Immigration Residuals

29

ln(Wage 25th%) ln(Wage 75th%) ln(Wage 25th%) ln(Wage 75th%)

L_Offshoring 0.044*** 0.010** 0.048*** 0.016***[0.005] [0.004] [0.005] [0.004]

L_Immigration ­0.001** ­0.001** ­0.002*** ­0.001**[0.000] [0.000] [0.001] [0.000]

H_Offshoring ­0.039*** ­0.025*** ­0.063*** ­0.050***[0.008] [0.008] [0.010] [0.009]

H_Immigration 0.004*** 0.003*** ­0.004 ­0.004*[0.001] [0.001] [0.003] [0.002]

Inshoring 0.027*** 0.038*** 0.048*** 0.058***[0.008] [0.008] [0.010] [0.009]

Age 0.002 0.003 0.005** 0.005***[0.002] [0.002] [0.002] [0.002]

Education 0.198*** 0.258*** 0.215*** 0.274***[0.008] [0.008] [0.010] [0.010]

Male 0.009*** 0.009*** 0.009*** 0.009***[0.000] [0.000] [0.000] [0.000]

Black 0.001* 0.001* 0.001*** 0.001***[0.000] [0.000] [0.000] [0.000]

Asian 0.001 0.004 0.017*** 0.020***[0.004] [0.004] [0.006] [0.006]

Hispanic ­0.002* ­0.001 ­0.001 0.000[0.001] [0.001] [0.001] [0.001]

Married 0.001 0.002*** 0.000 0.002***[0.001] [0.001] [0.001] [0.001]

Single ­0.005*** ­0.002*** ­0.005*** ­0.002***[0.001] [0.001] [0.001] [0.001]

Observations 4,704 4,704 4,704 4,704R­squared 0.961 0.966 0.960 0.965

Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions areweighted by the sample size and include year, state, industry, state*year, and industry*year fixed effects.

TABLE 1Impact of Offshoring and Immigration on Native Wages

OLS IV

30

L_Offshoring L_Immigration H_Offshoring H_Immigration

L_Offshoring IV 0.879*** ­0.105 0.070*** 0.253***[0.014] [0.106] [0.015] [0.080]

L_Immigration IV ­0.001* 0.899*** ­0.003*** ­0.221***[0.001] [0.035] [0.001] [0.016]

H_Offshoring IV ­0.053** ­0.625*** 0.547*** 0.455***[0.024] [0.157] [0.016] [0.108]

H_Immigration IV 0.002 0.083* 0.002 0.461***[0.002] [0.047] [0.002] [0.034]

Observations 4,704 4,704 4,704 4,704R­squared 1.000 0.958 1.000 0.941F­Stat, Instruments 2315 364 566 63

TABLE 2First Stage IV Regressions

Robust standard errors in brackets. * significant at 10%; ** significant at 5%; *** significant at 1%. All regressions areweighted by the sample size, include all control variables, and include year, state, industry, state*year, and industry*year fixedeffects.

31

ln(W

age

10th

%)

ln(W

age

20th

%)

ln(W

age

30th

%)

ln(W

age

40th

%)

ln(W

age

50th

%)

ln(W

age

60th

%)

ln(W

age

70th

%)

ln(W

age

80th

%)

L_O

ffsho

ring

0.05

8***

0.04

7***

0.03

7***

0.02

8***

0.02

9***

0.02

4***

0.01

7***

0.00

7[0

.007

][0

.005

][0

.004

][0

.004

][0

.004

][0

.003

][0

.004

][0

.005

]L_

Imm

igra

tion

­0.0

01*

­0.0

01*

­0.0

01**

*­0

.001

***

­0.0

01**

*­0

.001

*­0

.001

­0.0

01**

[0.0

01]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

H_O

ffsho

ring

­0.0

68**

*­0

.048

***

­0.0

35**

*­0

.035

***

­0.0

30**

*­0

.026

***

­0.0

26**

*­0

.039

***

[0.0

14]

[0.0

09]

[0.0

07]

[0.0

07]

[0.0

07]

[0.0

07]

[0.0

07]

[0.0

08]

H_I

mm

igra

tion

0.00

10.

005*

**0.

004*

**0.

004*

**0.

003*

**0.

003*

**0.

003*

**0.

005*

**[0

.002

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

]In

shor

ing

0.05

2***

0.03

4***

0.02

7***

0.03

4***

0.02

8***

0.02

8***

0.03

4***

0.05

5***

[0.0

15]

[0.0

09]

[0.0

08]

[0.0

07]

[0.0

07]

[0.0

07]

[0.0

08]

[0.0

11]

Age

0.00

30.

003

0.00

30.

002

0.00

20.

001

0.00

30.

003*

[0.0

04]

[0.0

03]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

Educ

atio

n0.

190*

**0.

193*

**0.

204*

**0.

214*

**0.

228*

**0.

237*

**0.

250*

**0.

273*

**[0

.014

][0

.009

][0

.008

][0

.007

][0

.007

][0

.007

][0

.007

][0

.010

]M

ale

0.00

9***

0.00

9***

0.00

8***

0.00

9***

0.00

9***

0.01

0***

0.00

9***

0.00

8***

[0.0

01]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

01]

Blac

k­0

.001

0.00

00.

001*

**0.

002*

**0.

002*

**0.

001*

**0.

001*

0.00

0[0

.001

][0

.000

][0

.000

][0

.000

][0

.000

][0

.000

][0

.000

][0

.000

]A

sian

0.00

60.

006

0.00

10.

000

0.00

10.

004

0.00

60.

003

[0.0

06]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

05]

His

pani

c­0

.002

­0.0

02**

­0.0

01­0

.001

*­0

.001

*­0

.002

**­0

.001

0.00

0[0

.002

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

]M

arrie

d0.

001

0.00

00.

001

0.00

1*0.

002*

**0.

002*

**0.

002*

**0.

002*

**[0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

]Si

ngle

­0.0

05**

*­0

.005

***

­0.0

04**

*­0

.004

***

­0.0

03**

*­0

.003

***

­0.0

02**

*­0

.001

*[0

.002

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

][0

.001

]

Obs

erva

tions

4,70

44,

704

4,70

44,

704

4,70

44,

704

4,70

44,

704

R­sq

uare

d0.

930

0.95

60.

965

0.96

90.

969

0.96

90.

967

0.95

9

Impa

ct o

f Offs

horin

g an

d Im

mig

ratio

n on

Nat

ive

Wag

e D

ecile

s (O

LS)

TABL

E 3

Robu

st s

tand

ard

erro

rs in

bra

cket

s. *

sig

nific

ant a

t 10%

; **

sign

ifica

nt a

t 5%

; ***

sig

nific

ant a

t 1%

. A

ll re

gres

sion

s ar

e w

eigh

ted

by th

e sa

mpl

e si

ze a

nd in

clud

e ye

ar, s

tate

, ind

ustry

,st

ate*

year

, and

indu

stry

*yea

r fixe

d ef

fect

s.

32

ln(W

age

10th

%)

ln(W

age

20th

%)

ln(W

age

30th

%)

ln(W

age

40th

%)

ln(W

age

50th

%)

ln(W

age

60th

%)

ln(W

age

70th

%)

ln(W

age

80th

%)

L_O

ffsho

ring

0.06

2***

0.05

1***

0.04

3***

0.03

4***

0.03

5***

0.03

1***

0.02

4***

0.01

4***

[0.0

07]

[0.0

05]

[0.0

05]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

04]

[0.0

04]

L_Im

mig

ratio

n­0

.002

**­0

.002

***

­0.0

03**

*­0

.002

***

­0.0

02**

*­0

.001

***

­0.0

01**

­0.0

01**

*[0

.001

][0

.001

][0

.001

][0

.000

][0

.000

][0

.000

][0

.000

][0

.001

]H

_Offs

horin

g­0

.104

***

­0.0

71**

*­0

.060

***

­0.0

63**

*­0

.058

***

­0.0

49**

*­0

.050

***

­0.0

55**

*[0

.016

][0

.011

][0

.009

][0

.008

][0

.008

][0

.008

][0

.009

][0

.011

]H

_Im

mig

ratio

n­0

.008

**­0

.005

*­0

.006

**­0

.004

*­0

.005

**­0

.006

***

­0.0

06**

­0.0

06*

[0.0

04]

[0.0

03]

[0.0

03]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

03]

[0.0

03]

Insh

orin

g0.

086*

**0.

054*

**0.

047*

**0.

057*

**0.

051*

**0.

045*

**0.

053*

**0.

065*

**[0

.017

][0

.011

][0

.010

][0

.009

][0

.009

][0

.008

][0

.009

][0

.011

]A

ge0.

006

0.00

6**

0.00

7***

0.00

5***

0.00

5***

0.00

5***

0.00

6***

0.00

7***

[0.0

04]

[0.0

03]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

[0.0

02]

Educ

atio

n0.

209*

**0.

213*

**0.

226*

**0.

230*

**0.

245*

**0.

257*

**0.

268*

**0.

297*

**[0

.016

][0

.011

][0

.010

][0

.009

][0

.009

][0

.009

][0

.010

][0

.016

]M

ale

0.00

9***

0.00

9***

0.00

9***

0.00

9***

0.01

0***

0.01

0***

0.01

0***

0.00

9***

[0.0

01]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

01]

Blac

k0.

000

0.00

10.

002*

**0.

002*

**0.

002*

**0.

002*

**0.

001*

**0.

001

[0.0

01]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

[0.0

00]

Asi

an0.

025*

**0.

024*

**0.

020*

**0.

016*

**0.

017*

**0.

022*

**0.

024*

**0.

025*

**[0

.009

][0

.006

][0

.006

][0

.005

][0

.005

][0

.005

][0

.006

][0

.008

]H

ispa

nic

­0.0

02­0

.002

0.00

0­0

.001

­0.0

01­0

.001

­0.0

010.

001

[0.0

02]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

Mar

ried

0.00

00.

000

0.00

00.

001

0.00

1**

0.00

1**

0.00

2***

0.00

1**

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

Sing

le­0

.005

***

­0.0

05**

*­0

.004

***

­0.0

04**

*­0

.003

***

­0.0

03**

*­0

.002

***

­0.0

01*

[0.0

02]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

[0.0

01]

Obs

erva

tions

4,70

44,

704

4,70

44,

704

4,70

44,

704

4,70

44,

704

R­sq

uare

d0.

929

0.95

50.

963

0.96

80.

968

0.96

80.

965

0.95

6

TABL

E 4

Impa

ct o

f Offs

horin

g an

d Im

mig

ratio

n on

Nat

ive

Wag

e D

ecile

s (IV

)

Robu

st s

tand

ard

erro

rs in

bra

cket

s. *

sig

nific

ant a

t 10%

; **

sign

ifica

nt a

t 5%

; ***

sig

nific

ant a

t 1%

. A

ll re

gres

sion

s ar

e w

eigh

ted

by th

e sa

mpl

e si

ze a

nd in

clud

e ye

ar, s

tate

,in

dust

ry, s

tate

*yea

r, an

d in

dust

ry*y

ear f

ixed

effe

cts.

33

34

A Model Appendix

A.1 Deriving Equation (7):

Totally differentiating the ratio of (4) to (5) gives:

daLaH

(dwΩs + dΩw

s −dswΩs2

)−aLdaH

a2H

(dwΩs + dΩw

s −dswΩs2

)=dL(1+I)H(1−J) + dIL

H(1−J)−L(1+I)dHH2(1−J)

+L(1+I)dJH(1−J)2

or:

aLaH

(aL − aH)(wΩs

) (w + Ω− s

)= L(1+I)

H(1−J)

(L+ dI

(1+I) − H + dJ(1−J)

)The first terms on each side cancel following from the ratio of (4) to (5) and since the

native factor supplies are fixed then L = H = 0. Therefore:

(aH − aL)(wΩs

) (s− w − Ω

)= dI

(1+I) + dJ(1−J)

or:

(7) σ(s− w − Ω

)= dI

(1+I) + dJ(1−J)

where the elasticity of substitution is defined as:

σ =d(aHaL

)/(aHaL

)d(wΩ

s )/(wΩs )

= (aH−aL)(wΩ/s)(w+Ω−s)(w+Ω−s) = (aH − aL)(wΩ/s)

35